Image Processing Solutions for Probe Card Manufacturing, Inspection and Optimization, PhD Sales Director
Contents Oxford Lasers and Guide Plates The Challenges of Measuring Guide Plates Performance of the Imaging System Fourier Analysis, a Possible Solution The Graphic Interface Conclusion 2
Oxford Lasers Turn-Key Advanced Laser Systems and Tools, for Industry and R&D Contract Laser Micromachining Services More than 1000 lasers / systems in 27 countries, headquarters in Oxfordshire (UK) and Boston (US) 15 years experience in vertical guide plate production High aspect ratio hole drilling Fine Cutting, Dicing, Scribing Selective Laser Patterning ProbeDrill Tool 3
Guide Plates for Probe Cards What are Guide Plates? Consist of a mechanically stable substrate with 1000 s of micro-holes, through which probes are fitted, ensuring accurate location of each probe A typical probe card uses several guide plates 62.5 X 60 µm 55 µm 70 µm entrance and exit Guide Plate Features Accurately locate probe pins Probe size and pitch to accurately match DUT Provide appropriate guiding/sliding of probes, scrub, wear, cleaning etc. Material is chosen accordingly (Si 3 N 4, Alumina, Macerite, Photoveel, Polyimides - Kapton, Vespel, Cirlex etc..) 4
Measuring Algorithm, Software, Graphic Interface SMART Project (Innovative UK, grant agreement 700495) Optics, Beam Delivery, Mechanical Design Drilling smaller, more precise holes, quicker Radiation - Matter Interaction, Laser Process Development 5
Measurement Challenges Non conventional features : for typical image processing software Niche market : no investment appeal for an industrial solution from large image processing software companies Lack of well defined standards A real need for a solution for quality control, optimization, customer support and R&D X-Y dimensions Entry rounding Corner radius of exit holes Position Hole taper Tailored, in-house development of an image processing solution 6
Defining the Imaging System Performance For several combinations of NA and pixel size, multiple pictures of the same sample have been taken. The standard deviations of each group of pictures defines the repeatability n Delta X Reference guide plate under best achievable conditions Delta Y High quality 70 µm round holes as an example Pitch between the micro holes centre of mass, is used as a relative measure of linear distance Samples have been cleaned Homogenous illumination conditions 7
Defining the Imaging System Performance For a given NA, we can determine the repeatability for a chosen pixel size System performance can be optimized with respect to hole dimensions After this study, we set the system to have a repeatability of +/- 50 nm for feature dimensions above 25 µm Abbe/diffraction limit: dd = λ 2 NNNN Pixels within Abbe limit: NN = dd pppppppppp ssssssss 8
Edge Detection, Sub-Pixel Resolution With Sub-pixel resolution we can extract more information from lower NA and larger field of views The software looks over several pixels along the contour to provide a smoother transition between adjacent pixels Before sub-pixel thresholding After sub-pixel thresholding Y [µm] 22 24 26 28 0.1 NA 0.25 NA 0.4 NA 0.75 NA 26 28 30 32 26 28 30 32 X [µm] X [µm] 9
50 X 60 µm square exit holes Edge Contour Overlap The edge contour overlap is an immediate and intuitive method to address process variability Example on the right: process with higher variability in Y with respect to X direction Author 10
Searching for the Best Algorithm Many different methods were studied during the SMART project : Holes dimensions, taper and position are easier to measure Corner radius and entry rounding are extremely challenging CASE STUDY: Corner radius measurement Software fitting of corner radius is much better than Operator fitting. But which is the best method/algorithm to use? Select a number of points and fit to a circle Relationship between largest inscribed square and smallest out-scribed square Measure distance at 45 degs These methods rely on arbitrary assumptions: Where does the corner start? Which points do I take for the fit? 11
Fourier Analysis rr θθ θθ Plotting the distance from the centre of mass to the edge as a function of the angle we can then extrapolate much information thru Fourier analysis By fitting the series with multiple terms we can evaluate corner radius, rotation, dimensions, parallelism etc... mm rr θθ = aa 00 + nn=11 aa nn cccccc nnnn + bb nn ssssss nnnn 12
30 X 30 µm square exit holes Fourier Analysis: Dimensions (a) Corner radius represented as the peak difference from ideal case (a) Dimension in X (b) and Y(c) are identified by the value at 0 and 90 deg. (c) (b) Raw Ideal case 13
Fourier Analysis: Hole Rotation 40 X 45 µm square exit holes rotated θ Rotation is equivalent to a shift along the θ axis. This method can easily deal with rotated features (even within the same image frame) shift Raw 10 terms Fit 14
Fourier Analysis: Debris Detection All the hole information is captured within the non-zero terms of the Fourier Series The method also allows for quick detection of residual debris Holes # mm rr θθ = aa 0 + aa nn cos nnnn + bb nn sin nnnn nn=1 15
Thresholding In order to be interrogated - the picture must be converted into different formats depending on the application (grey scale, binary etc ) A threshold based on intensity level is needed to perform all the operations Different thresholds [from 0 to 1] can give drastically different dimensions, especially when close to 1 or 0 What is the best threshold? Threshold [0-1] DimX [µm] DimY [µm] 0.075 61.68 52.30 0.100 61.42 52.03 0.150 61.13 51.73 0.200 60.95 51.54 0.250 60.82 51.40 0.300 60.70 51.26 0.350 60.60 51.14 0.400 60.49 51.02 0.450 60.40 50.91 0.500 60.32 50.79 0.550 60.21 50.67 0.600 60.12 50.53 0.650 60.00 50.40 0.700 59.87 50.24 0.750 59.72 50.00 0.800 58.60 48.27 16
Target dimensions 50 X 60 µm Thresholding and Illumination 1 2 3 4 5 6 7 8 9 Level Dim X [µm] Dim Y [µm] Saturation? Before saturation 1 60.08 50.65 No STD X [µm] 0.02 2 60.08 50.59 No STD Y [µm] 0.03 3 60.09 50.63 No 4 60.09 50.64 No 5 60.12 50.69 No 6 60.27 50.84 Yes 7 60.60 51.26 Yes 8 60.99 51.71 Yes 9 61.31 51.93 Yes Automatic thresholding provides stable results only when the image is not saturated Automatic illumination adjustment can avoid quality check failure due to operator error 17
Graphic Interface Software Demo 1. Load and analyse all the pictures within a folder 2. Automatic shape recognition (circular or square) - measures position, XY dimensions etc etc 3. Magnification selection with consequent automatic adjustment of the measured parameters 4. Selection of optional operations e.g. plotting hole labels, contour or best shape fit 5. View analysed images e.g. Silicon Nitride and Kapton 6. Selection of intensity format options: real colours, grey scale and binary 7. Data export into an excel file, saves position, dimension, corner radius etc etc 18
Graphic Interface 19
Graphic Interface Summary : Analysis of 150 square holes under same FOV (600 X 450 µm) in much less than 0.5 sec Corner radius, pitch and ellipticity (for circular holes) need to be added to the graphic version Valuable tool for both pre-production (fine adjustment of the drilling tool) and post-production phases (quality checks) 20
Future Work SHORT TERM Complete analysis and coding of the Fourier method Finalizing the graphic interface LONG TERM Automatic image grabbing Full mapping of the guide plate Real time analysis of samples during laser machining 21
Conclusion There is a real need for a standard measuring system/procedure for guide plates in the probe card community This in house developed, image processing system unlocks higher quality holes, rapid development/optimization and reduced time to market. System repeatability of +/- 50 nm has been achieved studying features from 25 to 70 µm Fourier analysis is a powerful and flexible method for measuring guide plates (dimension, corner radius, rotation, ellipticity etc ) Any imaging system will give you an output, but this output must be interpreted with a clear knowledge of the measuring system IF YOU CANNOT MEASURE IT, YOU CANNOT MAKE IT 22
Thank you for your time and attention 23